Paid search and shopping advertising for a store is the fastest channel to turn on and the easiest to run at a loss without noticing. The reason is that the numbers most agencies report, return on ad spend, conversion rate, click cost, all sit above the costs that decide whether a sale was worth making. A campaign at a healthy reported return can still be losing money once cost of goods, shipping and returns are subtracted, and the account will look fine for months. Add the fee model question, where a percentage of spend rewards spending more, and the buyer needs a clear view of both the pricing and the measurement before signing. This page covers both.
How the fee model shapes what you get
There are three usual structures. A flat monthly fee is the cleanest, because the agency earns the same whether your budget goes up or down, and it makes recommending a cut possible. A percentage of ad spend is the most common and quietly aligns the agency with growth in spend rather than in profit, which is not automatically wrong but should be recognised. Performance-based fees sound attractive and depend entirely on a metric both sides trust, which for a store with returns and repeat purchases is harder to define than it appears. Whichever model you choose, ask what the fee covers in hours, who does the work, and whether creative production and feed management are included or billed separately. Feed work in particular is where scope quietly disappears, since a shopping account is only as good as the product data feeding it.
The product feed is most of the job
For shopping campaigns the feed does the work that keywords do in search: titles, product types, attributes, images and pricing determine which queries an item can appear against and how it looks when it does. Improving titles and attributes usually moves performance more than bid adjustments do, and it is also the part of the account most often neglected because it is tedious and touches the store platform rather than the advertising interface. The same product data that feeds advertising should be exposed on your own pages with structured markup, since Google documents the product properties it supports for price, availability and returns, and consistency between the two prevents disapprovals. Ask a candidate agency who owns the feed, how often it updates, and how they handle out of stock items. A team that has not asked about your platform before quoting has not thought about the feed.
Measuring against margin, not revenue
Insist on reporting that subtracts cost of goods, shipping and expected returns before anything is called a return on spend. Categories with high return rates can invert an apparently healthy account entirely, and the agency will not usually volunteer that adjustment because the resulting numbers are less flattering. Agree the attribution window and the model at the start, and hold them fixed, because changing either mid-flight makes every historical comparison meaningless. Where a store runs both search and shopping alongside organic, be careful about branded search: it frequently reports superb returns while largely buying clicks you would have received free. Run a branded holdout for a fortnight and find out. Any performance claim used to sell you the service should be one you could verify yourself; advertising rules apply to agencies selling their own services too.
Setting up the engagement so it can be judged
Own the advertising accounts, the analytics property and the merchant centre yourself and grant the agency access, never the reverse, so the account history and the learning survive a change of provider. Agree what the first thirty days produce, since a well run start is usually feed and tracking cleanup rather than new campaigns. Ask for a named client at a comparable catalogue size and speak to the person who reviews the numbers weekly. Where paid sits alongside the wider store programme, keep the budgets and the reporting separate rather than folding them into a single ecommerce marketing retainer, because a blended figure hides which half is working. Then review at ninety days against margin, and be prepared to cut spend rather than raise it if the honest number says so.
Questions people ask about e-commerce ppc management
Is a percentage of spend fee ever the right model?
It can be, for a growing account where the work genuinely scales with budget, and it is simple to administer. The problem is the incentive at the margin: it makes recommending a reduction expensive for the agency. If you use it, cap it, review it annually, and make sure the contract does not penalise you for cutting spend. A flat fee with a defined scope avoids the conflict altogether.
Should we bid on our own brand name?
Test it rather than assume. Branded clicks are cheap and report excellent returns, which is why they flatter every account, but a share of them would have arrived through the organic listing at no cost. Pause branded campaigns in a controlled way for a fortnight and measure total branded orders, not just paid ones. Competitors bidding on your name change the answer, so retest periodically.
How much budget is needed to start?
Enough to gather data in a reasonable time, which depends on your product price and conversion rate rather than on a fixed figure. A store with a high order value needs fewer conversions to learn from. Ask the agency to work backwards: how many conversions per campaign per month do they need to make decisions, and what spend does that imply at your current rates. That is a defensible number.
What should the first month look like?
Mostly cleanup: conversion tracking verified against real orders, the product feed corrected, obvious waste cut, and account structure simplified before anything new launches. If a proposal opens with a large set of new campaigns and no tracking audit, the agency is building on numbers it has not checked. Ask for the tracking verification in writing before the first spend increase.